This study has two objectives: I) to extend observations on approximately 51,500 former students from Harvard and the University of Pennsylvania in order to identify characteristics of college youth and alumni years predictive of site-specific and total cancers later in life; II) to develop improved quantitative methods for detecting combinations of these predictive characteristics. Physical examination and other college records of the years 1916-1950 comprise the initial data to be examined for associations with fatal and nonfatal cancers identified on death certificates or on self-assessed mail questionnaires during a 27-62 year followup period. To date, approximately 1,350 cancer deaths have occurred among these former students in approximately 1.85 million person-years of observation. The number of surviving (nonfatal) cancer patients has yet to be assessed. The study affords a prospective perspective for both hypothesis testing and hypothesis seeking. Analytic methods will include multivariate case-control and exposed-nonexposed comparisons. The proposed study will utilize predocumented data on young adults and older alumni. Thus, unlike cross-sectional or retrospective studies, this approach avoids the need to infer previous from present behavior or to recall characteristics antecedent to cancer diagnosis.